Abstract: To address issues such as weak alignment between infrared and visible images, difficulties in small-target detection, and severe occlusion in complex backgrounds in aerial object detection, this study proposes an infrared-visible fused aerial object detection algorithm based on CFE-YOLOv11. Using YOLOv11 as the baseline, the proposed algorithm improves the CoDAFusion module at the fusion stage to achieve dynamic cross-modal alignment and complementary feature fusion through offset guidance and large-receptive-field contextual modeling; in the visible branch, a Frequency-Adaptive Dilated Convolution (FADC) module is introduced. Through frequency-domain decoupling and adaptive weighting, it enhances high-frequency texture features of small targets. In the infrared branch, an EBlock module is adopted to restore occluded-region features via global frequency-domain reconstruction and multi-scale compensation, improving robustness against occlusion in complex scenes. Experimental results demonstrate that CFEYOLOv11 achieves mAP@0.5 of 85.8% and 69.1% on the DroneVehicle and VEDAI datasets, representing improvements of 3.7% and 2.1% compared to the original YOLOv11 algorithm, indicating that the proposed algorithm can effectively improve the performance of UAV dual-light object detection.